1. The responses of 262 motor cortex cells and 223 cerebellar neurons were recorded during whole-arm reaching movements toward targets lights in eight evenly distributed directions radiating from a common central starting position. The reaching movements were followed by a 2-s target hold period where a fixed arm posture was actively maintained to stabilize the hand over the target light. 2. Cerebellar neurons had a higher mean tonic discharge rate while holding over the starting position (22.9 imp/s) than did motor cortex cells (12.5 imp/s). The mean population response curve describing the changes in activities with movement direction was likewise shifted toward higher frequencies in the cerebellum compared with the motor cortex, but the amplitude of the two curves was about equal. Therefore, the baseline discharges of cerebellar neurons were higher, but their changes in activity during movement were similar to those of motor cortical cells. 3. Motor cortex neurons were more strongly related to active maintenance of different arm postures than were cerebellar units. This was shown by a larger posture-related population response curve in the motor cortex (half-wave amplitude of cosine function was 11.2 imp/s, compared with 7.0 imp/s for cerebellar neurons), which represented the average response curve calculated from all the cells of the population. Furthermore, the motor cortex population had a higher percentage of single cells with tonic responses while the hand was held over different targets (tonic and phasic-tonic cells composed 57% of motor cortex population, compared with 38% of cerebellar population). Proportionately more cerebellar cells were phasically related to the movements. 4. The majority of motor cortex cells (58%) showed reciprocal changes relative to the center-hold time activity where the activity increased for movements in the preferred direction and decreased for movements in the opposite direction. Most of the remaining cells (40%) showed graded changes where the activity increased gradually as reaching was directed closer to the preferred direction. In contrast, the most common cerebellar response pattern was graded (38%). Only 26% were reciprocal and 18% were non-directional. The remaining 2% of motor cortical cells and 18% of cerebellar neurons could not be readily assigned to any of these three response classes. 5. Sector widths were calculated to measure the dispersion of individual cerebellar and motor cortical cell activities about the eight movement directions. Sector widths calculated from the absolute activities were always broader for cerebellar neurons (i.e., the cells were more broadly tuned).(ABSTRACT TRUNCATED AT 400 WORDS)